Christie’s October 2026 AI auction and the intellectual property divide

The Christie’s AI conflict

The Christie’s October 2026 AI-generated NFT auction follows the 2025 "Augmented Intelligence" auction that drew a letter from 3,500 artists. These artists claim that Christie’s supports AI models that use copyrighted work without license or payment to the original creators. Nicole Sales Giles, the director of digital art at Christie’s, stated that the training process constitutes influence rather than theft. This tension connects to Refik Anadol’s "Unsupervised" project at MoMA, which uses machine learning to process 138,151 pieces of MoMA metadata. Anadol’s work uses machine learning classification models to process datasets and produces images from 200 years of art. He works in dialogue with scholars like Jaakko Lehtinen of NVIDIA Research to develop these technologies. This project uses a 16-foot-wide screen that displays constantly changing patterns of colors, shapes, and textures, while the accompanying surround sound system produces ambient noises that fill the museum space. Ryan Zurrer donated the piece to MoMA through his 1OF1 Collection. The exhibition at MoMA, which debuted late last year, extended through October 29 due to its popularity.

Market trends and Sotheby’s holdings

Sotheby’s maintains a presence in the generative market through collections like the "Grails" series, which contains some of the most significant digital artworks ever assembled. This collection, seized from the bankrupt Three Arrows Capital, includes works such as Snowfro’s Chromie Squiggle #1780 and Tyler Hobbs’ Fidenza #725. While digital art continues to grow, NFT sales comprise 2% of the global art market. You know the difference between owning a token and owning the underlying intellectual property, but the market treats them differently.

Sotheby’s auctions include the "Grails: Starry Night" collection, which uses digital technology to reinterpret historical masterpieces. The auction house also sold a painting by the AI artist Ai-Da for over $1 million last year, following the $69.3 million Beeple sale conducted by Christie’s. Demand for AI-generated art is expected to increase over the next 12 months, as 67% of art enthusiasts predict that AI-generated art sales will pick up. 40% of art collectors anticipate that more people will buy these works. 28% of collectors have purchased AI-generated art and 52% say they expect to do so.

Legal realities and collector sentiment

The Christie’s October 2026 auction is the most direct test of whether collectors accept AI works despite the ongoing lawsuits regarding the use of copyrighted training data. Legal precedents create hurdles for AI-generated assets. The U.S. District Court of Columbia held in 2023 that AI systems cannot hold copyright for their own work because human authorship is a bedrock requirement. Judge Beryl A. Howell found the U.S. Copyright Office correctly denied copyright protection to works created without human involvement. This legal reality aligns with the concerns of many collectors. 82% of art collectors and 76% of enthusiasts want clearer distinctions between AI-generated art and man-made content. 69% of older collectors and 67% of younger collectors consider AI-generated art to be less important than human creativity.

Larva Labs once issued a DMCA takedown notice to OpenSea to preserve the value of its V2 CryptoPunks collection after buyers relisted V1 NFTs. This shows how immutable blockchain technology tests the practical realities of copyright enforcement. In Miramax, LLC v. Tarantino, the court is weighing whether a director owns the derivative rights to sell NFTs based on screenplay publication rights. Nike also filed a trademark infringement lawsuit against StockX because the marketplace sold NFTs encompassing the Nike mark. 66% of collectors bought from newly discovered artists in the past year. Only 16% of seasoned collectors believe that AI-generated art could reach the same price levels as traditional human-created art. Will the current legal focus on the source of training data dictate the future floor price of AI-generated works?